Storage rightsizing is matching each volume and dataset to the capacity and performance it actually needs, rather than the size and tier chosen at creation. With block storage you pay for what you provision, not what you fill, so a 1 TB volume holding 200 GB bills for the full terabyte every month. Add premium performance tiers nobody's workload requires and you are paying twice for headroom: once for empty space, once for unused speed.
This article is part of our complete guide to cloud storage and data cost optimization, the cluster pillar it links up to. Storage rightsizing is the storage-layer version of the Cut step in our See, Cut, Lock, Run method, and it begins with the inventory work in how to audit your cloud storage footprint.
Block volumes bill on allocated size regardless of how full they are. A fleet of half-empty volumes is pure waste. The first storage rightsizing move is simply finding the gap between provisioned and used capacity across the estate.
Step 1: Find the gap between provisioned and used
Inventory every block volume with two numbers: provisioned size and actual used size. The volumes where used sits far below provisioned are your candidates. This is the storage equivalent of reading utilization before resizing compute, the same discipline as in rightsizing compute. Pay special attention to volumes that were over-allocated at creation for fear of running out, since those rarely get revisited and the headroom is often enormous.
Step 2: Shrink or repack, carefully
Reducing a volume's size is not always a direct operation; several cloud block stores allow growing a volume in place but not shrinking it, so rightsizing down may mean creating a smaller volume and migrating the data, then cutting over during a maintenance window. Plan the migration like any data move: snapshot first, validate the copy, keep the rollback obvious. For new volumes, provision close to the real need with a sensible growth margin rather than defaulting to round numbers, because shrinking later is the costly path.
Want your storage rightsized for you?
Our cloud cost audit inventories every volume, flags the over-provisioned and over-tiered ones, and hands you a safe resize and migration plan ranked by dollars. On the performance model, you pay only from realized savings. No savings, no fee.
Book a cloud cost audit →Step 3: Match the performance tier to the workload
Capacity is only half the bill. Cloud block storage sells performance tiers, and the premium ones cost a multiple of standard storage. A volume on a high-performance tier that serves a low-IOPS workload is paying for speed it never uses. Read each volume's actual IOPS and throughput, then move workloads that do not need premium performance down to a standard tier. Where a provider lets you provision IOPS independently of capacity, tune the provisioned IOPS to measured demand rather than leaving it at a padded default.
Step 4: Tier cold data off expensive storage
Not all data deserves to sit on block or hot object storage. Data that is rarely read belongs on a cheaper tier, and data kept only for compliance belongs on archive storage. The systematic way to move data by how it is accessed is in how to build a storage lifecycle policy, and the decision of when archive pays off is in cold and archive storage: when it pays off. Combining capacity rightsizing with tiering is where the largest storage savings come from.
| Signal | What it means | Action |
|---|---|---|
| Used far below provisioned | Paying for empty space | Migrate to a smaller volume |
| Premium tier, low IOPS | Paying for unused performance | Move to a standard tier |
| Provisioned IOPS padded | Over-tuned at creation | Set IOPS to measured demand |
| Cold data on hot storage | Wrong tier for access pattern | Lifecycle to cool or archive |
Volume resize constraints, performance tiers, and independent IOPS provisioning differ across AWS, Azure, GCP and OCI and change. Verify current shrink support and tier options in each provider's documentation before migrating, as of May 2026.
Object storage: rightsizing means cleanup, not resizing
Block volumes have a provisioned size to shrink, but object storage bills purely on what you store, so rightsizing there means deleting and deduplicating rather than resizing. Three sources of object-storage bloat recur on almost every estate. Incomplete multipart uploads, the fragments of failed or abandoned large-object uploads, can accumulate into terabytes that never appear in a normal bucket view and bill silently until a lifecycle rule expires them. Object versioning, switched on for safety, keeps every prior version of every object forever unless a rule trims old versions, quietly multiplying the footprint of any frequently updated dataset. And plain duplication, the same files copied across buckets and accounts, is common wherever teams move fast. Enable a lifecycle rule to abort incomplete uploads after a few days, set version-expiry on versioned buckets, and run a periodic scan for duplication. The automation pattern for all of this is in how to build a storage lifecycle policy.
The Cloud Storage and Egress Cost Playbook includes the volume inventory queries and the scoring model we use to rank over-provisioned and over-tiered storage by dollars. It is the downloadable companion to this method.
Right-size at provisioning, not just after
The cheapest storage rightsizing is the kind you never have to do, because the volume was sized correctly the first time. Since shrinking a volume is harder than growing one, the default to provision generously is exactly backwards from a cost standpoint. Push the discipline left into how infrastructure is created: set sensible default sizes and performance tiers in your infrastructure-as-code modules, so a team that does not specify a size gets a reasonable one rather than a padded one, and require a justification for premium tiers rather than offering them by default. Add a policy check that flags new volumes provisioned far above any stated need before they ship. This turns rightsizing from a recurring cleanup into a property of how the platform is built, the Lock step of our method applied to storage. Combined with the lifecycle automation in how to build a storage lifecycle policy, it keeps the footprint right without a human chasing it every quarter.
The short version
Find the gap between provisioned and used capacity, shrink or repack carefully with a tested migration, match the performance tier and provisioned IOPS to real demand, and tier cold data off expensive storage. Start from the inventory in how to audit your cloud storage footprint. When you want it run across the whole estate at once, that is what our rightsizing and waste elimination service delivers.